The platelet-to-lymphocyte ratio as a predictor of patient outcomes in ovarian cancer: a meta-analysis

Abstract

Objectives: The platelet-to-lymphocyte ratio (PLR) is a predictive clinical biomarker for different cancers. However, the results of several studies investigating the association between the PLR and the prognosis of ovarian cancer have been inconclusive. Therefore, there is a need to conduct a meta-analysis to estimate the prognostic value of the PLR in ovarian cancer.

Methods: We searched the EMBASE, Medline, PubMed, and Web of Science databases to identify clinical studies that had evaluated the association between the PLR and ovarian cancer prognosis. Outcomes evaluated included overall survival (OS) and progression-free survival (PFS). We also analyzed PLR differences between malignant ovarian masses and the controls.

Results: Twelve relevant studies that comprised 2340 patients were selected for the meta-analysis. The results revealed that elevated PLR was significantly associated with poor OS (hazard ratio (HR) 1.63, 95% confidence interval (CI) 1.05–2.56, p < 0.01) and PFS (HR 1.61, 95% CI 1.03–2.51, p < 0.01). The PLRs in malignant cases were higher than in controls (mean difference = 63.57, 95% CI 39.47–87.66, p < 0.00001).

Conclusion: An elevated PLR is associated with poor prognosis in patients with ovarian cancer. The PLR could be employed as a prognostic marker in patients with ovarian cancer.

Keywords:

• Platelet-to-lymphocyte ratio
• ovarian cancer
• meta-analysis

Introduction

Ovarian cancer (OC) is the fifth leading cause of cancer-related mortality among gynecologic cancers; however, less than 40% of women with OC were diagnosed in the early stages of the disease. A primary cytoreductive surgery followed by adjuvant chemotherapy is the standard treatment for most patients¹. Even though most patients have an initial good response to chemotherapy, almost 75% of them experience a resurgence of the disease, with poor overall survival². On account of advances in chemotherapy and surgical techniques, the prognosis of OC has improved. However, the survival of OC patients is still not optimistic. Survival statistics show 5-year survival can change according to stage and therapy; in stage 1 it was about 90%, in stage 2 it was more than 40%, in stage 3 it was almost 20%, and around 5% in stage 4³. A major reason for this poor outcome is that early stages of this disease are usually asymptomatic and there are no specific diagnostic biomarkers for the patients at early disease stage.

The gold standard for differential diagnosis of adnexal masses is surgery⁴. Preoperatively, the standard tools for detecting OC are pelvic ultrasonography and measuring serum cancer antigen 125 levels, which could be combined with the menopausal status to calculate the risk malignancy index⁵. However, in daily practice, due to the performance limitations and time consumption of the standard tools, new prognostic assessment methods that are easy to use, inexpensive, and effective are needed for these patients. Research conducted in recent years has yielded many prospective biomarkers for predictors of OC prognosis, but they lack sensitivity, specificity, or reproducibility. Because of these drawbacks, they are still not accepted as part of the routine diagnostic algorithm for OC. Thus, the identification of accurate and reliable biomarkers and prognostic indicators is key for the improvement of patient outcomes and survival.

Recently, many studies have focused on hematological measurements or inflammatory factors, which are associated with the tumor microenvironment, for the diagnosis and prognosis of different cancers⁶. Previous clinical studies have confirmed that increases in the neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio (PLR), neutrophil counts, or platelet counts were associated with negative clinical pathologic characteristics, such as a higher risk of recurrence, aggressive tumor biology, and greater disease progression, in patients with various types of cancer^7,8. Platelets may release growth factors such as transforming growth factor β (TGF-β), platelet-derived growth factor (PDGF), and vascular endothelial growth factor (VEGF), which may contribute to the growth, progression, and metastasis of the tumor by stimulating the proliferation of ovarian tumor cells⁹. Furthermore, a reduced lymphocyte count is an indicator of a reduced immune response, and it is correlated with higher mortality in patients with OC than in healthy controls or patients with benign diseases¹⁰. The PLR, which is inexpensively and easily determined in routine clinical practice, was proposed to be a highly efficient and independent prognostic biomarker for many tumors⁷. Although there are several studies already published examining the relationship between PLR and cancer, the role of PLR in OC prognosis remains unclear¹¹.

In light of these findings, we performed a meta-analysis to determine the predictive value of the PLR for prognosis in OC, and to analyze the difference in PLR between patients with malignant ovarian masses and those with benign ones.

Methods

Search strategy

A comprehensive search strategy was employed to search the EMBASE via OvidSP, Medline via OvidSP, PubMed and Web of Science databases for all relevant literature published before March 25, 2017. The following keywords were used in accordance with the search strategy: (“ovarian neoplasms”) AND (“platelet-to-lymphocyte ratio” OR “platelet lymphocyte ratio” OR “platelet-lymphocyte ratio” OR “PLR”). The search was performed using both free text and MeSH terms. In addition, references in the related literature were manually searched to identify potentially relevant articles.

Inclusion and exclusion criteria

The full text of a selected article was further screened for eligibility, if the title and abstract met the following criteria according to either reviewer: (1) OC was diagnosed on the basis of histopathological examination; (2) the PLRs of patients were measured prior to treatment; (3) correlation of PLR with prognosis (OS and PFS) was reported; or (4) the PLR was reported in patients with benign and malignant ovarian masses. The exclusion criteria were as follows: (1) review articles, conference abstracts, editorial comments, letters, expert opinions, or case reports; (2) studies with insufficient data to perform the meta-analysis. Two reviewers independently screened the title and abstract search results.

Extraction and quality assessment of data

Two researchers collected data independently in accordance with predesigned tables, which included author, publication year, country, ethnicity, study period, sample size, disease stage, histology, treatment, cut-off value, PLR (high/low), endpoint (OS or PFS), PLR in patients with malignant ovarian masses and controls, and number of patients in each group.

The quality of the included studies was assessed independently by two investigators using the Newcastle-Ottawa Quality Assessment Scale (NOS). NOS scores >6 were considered to indicate high-quality studies. Disagreements between reviewer assessments were resolved via group discussion and consensus.

Statistical analysis

Meta-analysis was performed using RevMan5.2, which was recommended by the Cochrane Collaboration. Hazard ratios (HRs) and their 95% confidence intervals (CIs) were obtained directly from included studies for all the survival outcomes, and log (HR) and standard error were calculated accordingly; the pooled HR was calculated by the method of genetic inverse variance¹². Mean differences were calculated for continuous variables. The heterogeneity among the included studies was assessed using the χ² test, with a p-value <0.10 indicating a significant difference. The random effects model was used when heterogeneity existed, and subgroup analysis was conducted to investigate the source of the heterogeneity. A funnel plot was used to evaluate publication bias. The total effect was evaluated using the Z test, and p < 0.05 was used to identify significant effects¹³.

Results

Study selection and characteristics

The literature retrieval procedure is shown in Figure 1. The original search algorithm retrieved 125 studies. After further discussion and consideration of the retrieved articles, 12 studies published between 2011 and 2016, involving 2340 patients, were selected for this meta-analysis. The study characteristics are summarized in Table 1. Notably, six studies were conducted in Turkey, two in China, two in Thailand, one in Poland, and one in the UK. The NOS scores, which indicated the quality of these studies, ranged from 6 to 8.

Figure 1. Flow diagram of included studies. PLR, platelet-to-lymphocyte ratio; OS, overall survival; PFS, progression-free survival.

Table 1. Characteristics of included studies. Data for age are given as mean ± standard deviation (range).

Study	Publicationyear	Country	Ethnicity	Study period	Age (years)	Sample size	Platelet-to-lymphocyteratio cut-off value	NOS	Outcome
Asher^14	2011	UK	Caucasian	1988–1998	62 (24–90)	235	300	7	A
Zhang^15	2015	China	Asian	2000–2012	50.6 ± 11.1 (24–76)	190	203	8	A, B
Miao^16	2016	China	Asian	2005–2010	55 (45–84)	344	207	6	A, B
Raungkaewmanee^17	2012	Thailand	Asian	2004–2010	53 (23–85)	166	200	7	A, B
Badora-Rybicka^18	2016	Poland	Europeoid	2007–2013	54 (22–77)	315	129	7	A, B
Supoken^19	2014	Thailand	Asian	2003–2013	52 (45–58.8)	36	300	6	B
Topcu^20	2014	Turkey	Caucasian	2011–2014	53.5 ± 9.50	94	NR	6	C
Yildirim^21	2014	Turkey	Caucasian	2007–2013	49.5	299	NR	7	C
Ozaksit^22	2015	Turkey	Caucasian	2010–2013	17.7 ± 2.2 (10–21)	133	NR	6	C
Bakacak^23	2016	Turkey	Caucasian	2007–2015	62.00 ± 13.00	218	NR	7	C
Polat^24	2016	Turkey	Caucasian	2013–2015	39.4 ± 14.3 (17–82)	197	NR	6	C
Kemal^25	2014	Turkey	Caucasian	2008–2012	52.1 (23–88)	113	NR	6	C

NR, not reported; NOS, Newcastle-Ottawa Quality Assessment Scale; A, platelet-to-lymphocyte ratio and overall survival; B, platelet-to-lymphocyte ratio and progression-free survival; C, mean difference of platelet-to-lymphocyte ratio between malignant ovarian masses and controls.

Five of the selected eligible studies presented data regarding the association between the PLR and OS^14–18, five studies discussed the link between PLR and PFS^15–19, and six examined numerical differences in PLR between patients with malignant ovarian masses and the controls^20–25. The disease characteristics in 12 studies are summarized in Table 2.

Table 2. Characteristics of disease.

					Overall survival (months)		Progression-free survival (months)
Study	Stage	Histology	Treatment	PLR cut-off value	PLR-high	PLR-low	PLR-high	PLR-low
Asher^14	I–IV	Serous, mucinous, endometrioid, clear cell, undifferentiated, mixed type malignancy	Surgery, chemotherapy	300	14.5	37.4	NR	NR
Zhang^15	I–IV	Serous, mucinous, endometrioid, clear cell, undifferentiated, mixed type malignancy	Surgery, chemotherapy	203	28	64	11	24
Miao^16	I–IV	Serous, non-serous malignancy	Surgery, chemotherapy	207	3-year overallsurvival 21.3%	3-year overallsurvival 67.2%	11	25
Raungkaewmanee^17	I–IV	Serous, mucinous, endometrioid, clear cell, undifferentiated, mixed type malignancy	Surgery	200	21.8	60.7	11.5	19.4
Badora-Rybicka^18	I–IV	Epithelial ovarian cancer	Surgery, chemotherapy	129	NR	NR	NR	NR
Supoken^19	I–IV	Cystic, adenofibromatous, indeterminate type malignancy	Chemotherapy	300	NR	NR	10	34
Topcu^20	NR	Benign, malignant	Surgery	NR	NR	NR	NR	NR
Yildirim^21	NR	Benign, malignant	Surgery	NR	NR	NR	NR	NR
Ozaksit^22	NR	Benign, malignant	Surgery	NR	NR	NR	NR	NR
Bakacak^23	NR	Benign, malignant	Surgery	NR	NR	NR	NR	NR
Polat^24	NR	Benign, malignant	Surgery	NR	NR	NR	NR	NR
Kemal^25	NR	Epithelial ovarian cancer, healthy control	Surgery	NR	NR	NR	NR	NR

PLR, platelet-to-lymphocyte ratio; NR, not reported.

PLR and OS in ovarian cancer

Five studies that comprised 1250 OC patients reported HRs for OS. The results of the analysis of these studies are displayed in Figure 2^14–18. Our pooled results favored the low PLR patients (HR 1.63, 95% CI 1.05–2.56, p < 0.01), which indicated that a higher PLR was more strongly associated with poor OS. Heterogeneity was observed among these studies (I² = 93%, p < 0.01). Subgroup analyses were performed to explore the source of heterogeneity; when studies were stratified by cut-off value ≤200 and cut-off value >200, the data showed that no heterogeneity was found (I² = 0%, p = 0.76; I² = 26%, p = 0.25, respectively) (Figure 2). Subgroup analyses indicated that the cut-off value may be the source of heterogeneity.

Figure 2. Forest plots of studies evaluating hazard ratio with 95% confidence interval (CI) of platelet-to-lymphocyte ratios for overall survival.

PLR and PFS in ovarian cancer

Five studies comprising 1051 patients investigated the association between PLR and PFS^15–19. High PLR indicated an inferior PFS outcome, with a combined HR of 1.61 (95% CI 1.03–2.51, p < 0.01) (Figure 3), which indicated that patients with a higher PLR had a higher risk of disease progression than those with a low PLR. Heterogeneity among these studies was observed (I² = 89%, p < 0.01). When studies were subgrouped by cut-off value ≤200 and cut-off value >200, the data showed that no heterogeneity was found (I² = 0%, p = 0.53; I² = 56%, p = 0.13, respectively) (Figure 3). Subgroup analyses indicated that the cut-off value may be the source of heterogeneity.

Figure 3. Forest plots of studies evaluating hazard ratio with 95% confidence interval (CI) of platelet-to-lymphocyte ratios for progression-free survival.

Numerical differences in PLR between patients with malignant and benign ovarian masses

Six studies comprising 1054 patients reported the PLR in patients with malignant ovarian masses and the controls^20–25. A comparison of the malignant cases with the controls revealed a significant difference in PLR; the pooled mean difference was 63.57 (95% CI 39.47–87.66, p < 0.00001). Heterogeneity was observed (I² = 81%, p < 0.0001). Subgroup analysis stratified the controls by benign masses and healthy tissue; the pooled data are shown in Figure 4. The result indicated that the difference in controls may be the source of heterogeneity. The PLRs in the patients with malignant masses were higher than those in patients with benign masses and healthy tissue, indicating that patients with higher PLR tend to have a higher risk of OC.

Figure 4. Forest plots of studies evaluating mean differences of platelet-to-lymphocyte ratios in malignant ovarian masses and controls. 95% CI, 95% confidence interval.

Publication bias

Publication bias was evaluated by the funnel plot; Figure 5 indicates that potential publication bias may exist for OS and PFS. However, due to the small number of included studies, the funnel plots may not be significant.

Figure 5. Funnel plots of the meta-analysis of the impact of platelet-to-lymphocyte ratios on overall survival (A), progression-free survival (B) and mean difference between malignant ovarian masses and controls (C). SE, standard error; MD, mean difference.

Discussion

In this meta-analysis, by integrating the results of independent studies, elevated PLR was found to be associated with poor OS and PFS. PLRs in patients with malignant ovarian masses were higher than those in patients with benign masses. Heterogeneity was observed among the included studies; we tried to explore the source of heterogeneity by subgroup analyses. In the OS and PFS analyses, the cut-off value of PLR was the source of heterogeneity, and a higher PLR cut-off value had a higher clinical prognostic value. In the analyses of mean differences in PLR between malignant OC and controls, the control group was stratified by benign masses and healthy tissue; the difference in controls may be the source of heterogeneity.

A systematic review was conducted in 2014 to investigate the prognostic value of PLR in various cancers, with only two articles on OC²⁶. In addition, the interaction between clinical pathologic characteristics and the PLR has been illustrated in a variety of studies. Huang and colleagues reported that OC patients with a lower PLR showed a better treatment response to dose-dense chemotherapy. Thus, the PLR might be an indicator of treatment response in dose-dense chemotherapy, and can help clinicians to decide when to switch patients to another chemotherapy regimen²⁷. A study by Xiang and colleagues compared the MLR together with PLR between OC patients and controls; significant differences were observed²⁸. In a study by Kokcu and colleagues, PLR increased progressively with increases in the stage of OC, indicating that it could function as an independent prognostic factor related to the cancer stage²⁹. A study by Ashrafganjoei and colleagues reported an association between the PLR and surgical outcomes in patients with OC. High PLRs were associated with poor surgical outcomes and suboptimal debulking³⁰. Because the results of the above studies do not match the outcome of our analysis, they have not been included in our study. However, these findings have strongly supported the prognostic value of PLR in OC.

Study limitations include the insufficient number of studies as well as their retrospective nature, which might reduce the generalizability of the results. Second, the funnel plot suggests potential publication bias. Third, several studies have evaluated PLR in combination with other biomarkers, which may be of practical value to the clinic, but the number of studies cannot support meta-analysis. In the future, well-designed and high-quality clinical studies with a larger sample size are needed to draw definitive conclusions regarding the prognostic value of PLR in OC.

The following reasons may explain why PLR could serve as an accurate and reliable prognostic marker. First, PLR can be measured in routine blood tests performed in daily clinical practice, making it easy to measure, low cost, and reproducible^31,32. Second, the prognostic value of the PLR has been demonstrated in a wide variety of tumor types³³. However, the mechanisms by which tumors alter the PLR in patients with OC are still incompletely understood. Increases in the PLR may be linked to an increased platelet-dependent systemic inflammatory response together with a lower lymphocyte-mediated antitumor immune response, which reflect a supportive tumor microenvironment.

Platelets play major roles in hemostasis, thrombosis, inflammation, and vascular biology³⁴. In both mouse models and patients with OC, tumors induce thrombocytosis and thrombosis by stimulating IL-6 release and hepatic thrombopoietin expression^35,36. Recently, a high prevalence of platelet-derived venous thromboembolism has been reported in patients with cancer. Alternatively, platelets recruited by tumor cells may act as catalysts for acceleration of tumor growth, angiogenesis, and metastasis³⁷. When the tumor volume exceeds a certain size (> 1–2 mm³), angiogenesis occurs in tumors from pre-existing blood vessels, and provides the oxygen and nutrients essential to sustain tumor growth³⁸. Experimental evidence indicates that the stimulatory effect of platelets on angiogenesis may facilitate tumor growth and metastasis. Upon activation, platelets can release a series of pro-angiogenic factors such as VEGF, PDGF, basic fibroblast growth factor (β-FGF), insulin-like growth factor 1, lysophosphatidic acid, angiopoietin, matrix metalloproteinase (MMP)-1, MMP-2, and MMP-9³⁹. These signaling ways induces angiogenesis by increasing sprouting, vessel dilation, and the detachment of pericytes from pre-existing vessels, and the subsequent increase in blood supply potently facilitates rapid tumor growth and metastasis^40,41. These findings make the angiogenic growth factor system an attractive oncologic therapeutic target⁴². In addition, experimental data from several clinical trials suggest the possible benefits of anti-platelet drugs in limiting tumor initiation and progression. Many novel VEGF/VEGFR-targeted inhibitors have been reported to improve the prognosis and survival of patients more than chemotherapy alone does⁴², such as bevacizumab and ramucirumab^43–45. However, when VEGF signaling is blocked, other compensatory angiogenic signaling pathways can regulate tumor angiogenesis, including the PDGF, FGF, and hepatocyte growth factor/Met signaling pathways⁴⁶. Consequently, multi-agent therapy appears to be the optimal strategy for targeting angiogenesis. Genetic and molecular studies have led to a better understanding of biological characteristics of tumors, resulting in the development of multiple molecular targeted agents. These agents, which include sorafenib, sunitinib, axitinib, and pazopanib, aim to neutralize VEGF, PDGF, FGF, and other signaling pathways^47–50. Because OC is recognized as an extremely vascularized tumor, antiangiogenic agents have become a key focus of treatment.

In a variety of tumors, cancer cells often disrupt immune system homeostasis to their advantage in order to escape from immune surveillance. Indeed, an immune-suppressive milieu is a common in the tumor microenvironment. Appropriate immune responses, which have been intensively studied in recent decades, play an important role in controlling cancer growth^51,52. Lymphocytes are key mediators of immune surveillance and immune editing, and lymphocyte infiltration into the tumor microenvironment is a prerequisite for an immunologic antitumor reaction. High lymphocyte counts and infiltration of tumors by type I lymphocytes with greater dimensions are associated with a good prognosis in cancers such as colon, ovarian, lung, and breast cancer^51–53. Generally, a low lymphocyte amount could result in a weak and insufficient immunologic reaction to the tumor. A randomized clinical trial involving more than 1000 patients revealed that, in patients receiving neoadjuvant chemotherapy, the broad infiltration of lymphocytes into tumors seemed to be more clinically beneficial than no lymphocytic infiltration, in terms of achieving a pathologic complete response after chemotherapy⁵⁴.

In conclusion, this study shows that elevated PLR is closely associated with poor survival outcome in patients with OC. Patients with malignant ovarian mass have higher PLR than those with benign ovarian masses. Thus, the PLR represents a predictive biomarker with great clinical utility in patients with OC.

Conflict of interest

The authors report no conflicts of interest in this work.

Source of funding

This study was supported by National Natural Science Foundation of China Grant Number [81473643], Beijing Municipal Administration of Hospitals’ Youth Programme [QML20150903], Basic-clinical cooperation Science Foundation of Capital Medical University [15JL79].